Rigol DG4xxx ppulse and npulse

[Rick]

Has any owner of Rigol DG4xxxx experienced any problem with ppulse and npulse built-in arbitrary waveforms above 5 Mhz?
This is the problem:
Above approximately 5 Mhz the amplitude is not stable and the frequency is wrong and sometimes the signal is missing.

[Pinkus]

I can confirm the same behaviour with my DG4062.
Try to enter 10000000.01 Hz : you will see pulses for a second and then no pulses for a second.
Even worse if you enter 10000000.001 Hz.

[Uup]

I just checked my DG4162 and it also behaves the same as yours. Although, I found that the missing signal at 10MHz and 20MHz was intermittent when adjusting the frequency with the knob.

Definately a firmware bug and it should be reported Rigol.

Software version of mine is:  00.01.07.00.03
FPGA: 00.01.08.03
Hardware: 01.03
Keyboard: 04.01
Enc. FPGA: 06

[Rick]

I can confirm the same behaviour with mine DG4062.
Try to enter 10000000.01 Hz : you will see pulses for a second and then no pulses for a second.
Even worse if you enter 10000000.001 Hz.

Right:)
Thank you for the confirmation.

[Rick]

I just checked my DG4162 and it also behaves the same as yours. Although, I found that the missing signal at 10MHz and 20MHz was intermittent when adjusting the frequency with the knob.

Definately a firmware bug and it should be reported Rigol.

Software version of mine is:  00.01.07.00.03
FPGA: 00.01.08.03
Hardware: 01.03
Keyboard: 04.01
Enc. FPGA: 06

If it is a FW bug it is nothing serious. They will fix it.
My FW version is: 00.01.07
Hardware version: 01.03
Thank you.

[drieg]

Yes, Rigol just confirmed it is a bug, I'll post here when they come up with the fix.

[ted572]

Yes, Rigol just confirmed it is a bug, I'll post here when they come up with the fix.

Its been a Month since this issue was confirmed by Rigol as requiring a fix. Have you received any information about when it is expected to be resolved with the release of new FW to replace Version 00.01.07.00.03 ?

Thank you for any information you may have on the status.

[ted572]

DG4000 Calibration Restoration:
Re. Sniffing the Rigol's internal I2C bus, Reply #1771

I now believe that the Restoration of the DG4000 calibration should be done for AC, LFLAT, and HFLAT to correct some other potential glitches.  Nothing is being changed, or has to be connected to the DG4000, so the process goes fast.  Just bring up and save all the default values for each step, starting with 1, or 1-1 (A/R).
You can also do this for 'Inner Imped' and 'Offset', although I haven't seen a case or benefit for it yet.

I just wouldn't do this for Freq Accuracy, or Counter, unless you see a need for it, and then you should complete this cal. process as prompted.

The calibration restoration is effective for both channels, so it isn't necessary to repeat each cal. routine for the other channel. 

[W7NGA]

I am not familiar with your oscilloscope but insure that you have terminated the input in 50-ohms in your next testing, else you will be dealing with transmission-line artifacts such as standing waves and varying source loading at the generator. -cheers

[Rick]

I am not familiar with your oscilloscope but insure that you have terminated the input in 50-ohms in your next testing, else you will be dealing with transmission-line artifacts such as standing waves and varying source loading at the generator. -cheers

Right, that oscillloscope has 1 Mohm inputs. Thanks.

[ted572]

Please:  Are the DG4000's 'ppulse and npulse waveforms' above 5 Mhz still considered an issue?

[Teneyes]

Please:  Are the DG4000's 'ppulse and npulse waveforms' above 5 Mhz still considered an issue?

Yes still a Bug (Characteristic of DDS?) to watch for, with FW 1.07
See pics, of:
   10000000.01Hz
   10000000.001Hz 
   10000000.00 Hz 
    9999999.9  Hz 
Note Time base

[Rick]

Latest update on this: I received an email from Rigol after I personnally contacted them through their website. Here is their answer:

"... it's not a bug since ppulse in build-in
waveform and the points length is fixed ,DG4 use DDS theory to create
waveform .

DDS theory is using fixed sampling rate when change the frequency will jump
some points ,so if the jumped points is high level will make ppulse signal
seems losing some points .

This is a common problem of DDS generator ."

...

[jpb]

Latest update on this: I received an email from Rigol after I personnally contacted them through their website. Here is their answer:

"... it's not a bug since ppulse in build-in
waveform and the points length is fixed ,DG4 use DDS theory to create
waveform .

DDS theory is using fixed sampling rate when change the frequency will jump
some points ,so if the jumped points is high level will make ppulse signal
seems losing some points .

This is a common problem of DDS generator ."

...

I think the issue is that ppulse and npulse are defined as a single positive or negative value out of the total 16k of points. Since you don't enter a pulse length they are presumably defined as being very short. This means the sampling can miss them all together.

The wave form memory is 16K or 16,384 points. The fixed sample rate is 500MS/s or 2nsecs per sample. At 10MHz the period is 100nsecs or only 50 samples. This means it is sampling the 16K of memory every 16,384/50 or 327.68 points on average.

Most of the time it will sample every 327 points and sometimes it will sample every 328 points.

With most wave forms this won't matter much, maybe a bit of jitter (hence the 500psec jitter on square waves).

With npulse and ppulse the pulse bit of the waveform appears to be only a few points wide (they want the pulse to be as short as possible) so the jitter takes you on and off the pulse completely.

If you want narrow pulses which aren't missed, then you can define them yourself and use step-by-step mode to ensure every sample is used - of course you'll then not be able to get exact frequencies such as 10.01MHz, the resolution will be 2 nsecs.